Ignite 2020 Neural TTS updates: new language support, more voices and flexible deployment options

by Scott Muniz | Sep 22, 2020 | Uncategorized

This article is contributed. See the original author and article here.

Ignite 2020 Neural Text-to-Speech updates: new language support, more voices and flexible deployment options

This post was co-authored by Garfield He, Melinda Ma, Yueying Liu and Yinhe Wei  

Neural Text to Speech (Neural TTS), a powerful speech synthesis capability of Cognitive Services on Azure, enables you to convert text to lifelike speech which is close to human-parity.  Since its launch, we have seen it widely adopted in a variety of scenarios by many Azure customers, from voice assistants to audio content creation. We continue to push the envelope to enable more developers to add natural-sounding voices to their applications and solutions.

Today, we are happy to announce a series of updates to Neural TTS that extends its reach globally and allows developers to deploy it anywhere the data resides. This includes new languages available, new voices with rich personas, and on-prem deployment through docker containers.

18 new languages/locales supported

Neural TTS has now been extended to support 18 new languages/locales. They are Bulgarian, Czech, German (Austria),  German (Switzerland), Greek, English (Ireland), French (Switzerland), Hebrew, Croatian, Hungarian, Indonesian, Malay, Romanian, Slovak, Slovenian, Tamil, Telugu and Vietnamese. 

You can hear samples of these voices below.

With these new voices, Microsoft Azure Neural TTS supports 49 languages/locales in total.

14 additional voices released to enrich the variety

Customers use TTS for different scenarios and their requirements for voice personas can vary. To provide more options to developers, we continue to create more voices in each language. Besides the extension to support new locales, we’ve announced 14 new voices to enrich the variety in the existing languages.

Hear samples of these voices below.

With these updates, Microsoft Azure Text-to-Speech service offers 68 neural voices. Across standard and neural TTS capabilities, we now offer 140+ voices in total. Check the 70+ standard voices here.

More than 15 speaking styles available in en-US and zh-CN voices

Today, we’re building upon our Neural TTS capabilities in English (US) and Chinese (CN) with new voice styles. By default, the Text-to-Speech service synthesizes text using a neutral speaking style. With neural voices, you can adjust the speaking style to express different emotions like cheerfulness, empathy, and calm, or optimize the voice for different scenarios like customer service, newscasting and voice assistant that fit your need.

With the English (US) new voice, Jenny, which is created with a friendly, warm and comforting voice persona focusing on conversational scenarios, we provide additional speaking styles including chatbot, customer service, and assistant.

You can hear the different speaking styles in Jenny’s voice below:

A new speaking style is also available for the en-US male voice, Guy.  Guy’s newscast style can be a great choice for a male voice that can read professional and news related content. 

In addition, 10 new speaking styles are available with our zh-CN voice, Xiaoxiao. These new styles are optimized for audio content creators and intelligent bot developers to create more engaging interactive audios that express rich emotions.  

You can hear the new speaking styles in Xiaoxiao’s voice below:

For the Chinese voice Xiaoxiao, the intensity (‘style degree’) of speaking style can be further adjusted to better fit your use case. You can specify a stronger or softer style with ‘style degree’ to make the speech more expressive or subdued.

The style degree can be adjusted from 0.01 to 2 inclusive. The default value is 1 which means the predefined style intensity will be applied. The minimum unit is 0.01, which softens the style with a flatter tone. The value of 2 is the highest, which makes the style intensity obviously stronger than the default.

The SSML snippet below illustrates how the ‘style degree’ attribute is used to change the intensity of a speaking style.

The ‘style degree’ feature currently only applies to the Chinese voice Xiaoxiao and will come to more languages and voices later soon.

Check SSML for the details on how to use these speaking styles, together with other rich voice tuning capabilities.

Neural TTS Container is in public preview with 16 voices available in 14 languages

We have launched Neural TTS Container in public preview, as we are seeing a clear trend towards a future powered by the intelligent cloud and intelligent edge. With Neural TTS Container, developers can run speech synthesis with the most natural digital voices in their own environment for specific security and data governance requirements. Their Speech apps are portable and scalable with greater consistency whether they run on the edge or in Azure.

Currently 14 languages/locales are supported with 16 voices in Neural TTS Containers, as listed below. 

To get started, fill out and submit the request form to request access to the container. Currently Neural TTS Containers are gated and only approved for enterprises (EA customers) and Microsoft partners, and to an extent only for qualified customers.

Azure Cognitive Services Containers including Neural TTS Containers aren’t licensed to run without being connected to the metering / billing endpoint. You must enable the containers to communicate billing information with the billing endpoint at all times. Cognitive Services containers don’t send customer data, such as the image or text that’s being analyzed, to Microsoft. Queries to the container are billed at the pricing tier of the Azure resource that’s used for the ApiKey.

Here are the steps of how to install and run the container:

1. Make sure your machine to host the container meets the hardware requirements.
2. Get the container image with docker pull. For all the supported locales and corresponding voices of the neural text-to-speech container, please see Neural Text-to-speech image tags.
3. Run the container with docker run.
4. Validate that the container is running.
5. Query the container’s endpoint. Take AriaNeural voice for example, you can run below HTTP post method to get the TTS output audio:

Learn more about Container support in Cognitive Services and visit the Frequently Asked Questions on Azure Cognitive Services Containers.    

Get started

With these updates, we’re excited to be powering natural and intuitive voice experiences for more customers globally with flexible deployment options. For more information, visit below. 

• Try the TTS demo
• See our documentation
• Check out our sample code
• Learn about Speech containers

Step By Step: How to use SQLPackage with Access Token

by Scott Muniz | Sep 22, 2020 | Uncategorized

This article is contributed. See the original author and article here.

Purpose: 

SQLPackage allows you to authenticate with Access Token instead of providing Login name and password.

This article will show you how to do that end to end. 

General steps:

1. Create App Registration in your Azure Active Directory (AAD)
2. Create user for the Application to access Azure SQL DB and grant the needed permissions. 
3. Generate Access token for your Application. 
4. Use the Access token to import or export your database.

Detailed steps: 

1. Create App Registration in your Azure Active Directory (AAD)
1. Open Azure portal and access you Azure Active Directory management blade
2. Click on App Registrations
3. Click on New Registration
4. Give your application a name so it can be identified afterwards
5. Click on “Register”
6. Once the App is created you will be redirected to the App blade
7. Note your application (client) ID – you will use that later
8. Click on “Endpoints” at the top and note the “OAuth 2.0 token endpoint (v2)” url – we will use this later as well.
9. Click on “Certificate & Secrets”
10. Click on “New Client Secret”
11. Set the expiry time and click “Add”
12. Note the value of the key – we will use it later.
2. Create user for the Application to access Azure SQL DB and grant the needed permissions. 
   1. CREATE USER [SQLAccess] FROM EXTERNAL PROVIDER
   1. alter role dbmanager add member [SQLAccess]
1. Make sure your server has AAD Admin account configured.
2. Connect to you SQL DB with your AAD account
3. Create the user for the application access
4. Grant the needed permissions.
3. Generate Access token for your Application. 
1. Using PowerShell

$key= ConvertTo-SecureString `
         -String "{Key Secret}" `
         -AsPlainText `
         -Force
       
Get-AdalToken `
    -Resource "https://database.windows.net/" `
    -ClientId "{Application ID}" `
    -ClientSecret $key `
    -TenantId "{Tenant ID}"

1. Using C#

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
using Microsoft.IdentityModel.Clients.ActiveDirectory;

namespace ConsoleApp1
{
    class Program
    {
        static void Main(string[] args)
        {

            string clientId = "{Client ID}";
            string aadTenantId = "{Tenant ID}";
            string clientSecretKey = "{Key Secret}";

            string AadInstance = "https://login.windows.net/{0}";

            string ResourceId = "https://database.windows.net/";

            AuthenticationContext authenticationContext = new AuthenticationContext(string.Format(AadInstance, aadTenantId));

            ClientCredential clientCredential = new ClientCredential(clientId, clientSecretKey);
            DateTime startTime = DateTime.Now;
            Console.WriteLine("Time " + String.Format("{0:mm:ss.fff}", startTime));
            AuthenticationResult authenticationResult = authenticationContext.AcquireTokenAsync(ResourceId, clientCredential).Result;
            DateTime endTime = DateTime.Now;
            Console.WriteLine("Got token at " + String.Format("{0:mm:ss.fff}", endTime));
            Console.WriteLine("Total time to get token in milliseconds " + (endTime - startTime).TotalMilliseconds);
            Console.WriteLine(authenticationResult.AccessToken.ToString());
            Console.ReadKey();
        }
    }
}

4. Use the Access token to import or export your database.

1. Use your SQLPackage command and instead of using Login / User and password use the /AccessToken:{AccessTokenHere} (or /at)

Scale database up or down when geo replication is in use

by Scott Muniz | Sep 22, 2020 | Uncategorized

This article is contributed. See the original author and article here.